Search site

Using probabilistic data in the building industry

Objectives: This project, known as PROMETHEUS, aims to understand the needs of and hurdles faced by practitioners when using probabilistic climate data. The project uses UKCP09 data to develop a methodology for the creation of probabilistic future weather years compatible with common building simulation software. It also develops a methodology for estimating wind direction and speed, and shows the advantages of probabilistic data for the building industry.

UKCP09 products used

The Weather Generator 

How were UKCP09 products used?

1. The outputs of the UKCP09 Weather Generator were used to devise a methodology for the creation of probabilistic future weather files. 100 samples of 30 years of hourly data for each emissions scenario, time period and location were downloaded from the User Interface.

2. These were used to generate 100 test reference years (TRY) and 100 design summer year (DSY) files, one from each sample of 30 years. For both sets, the months were ordered by ascending mean air temperature and combined with like percentiles (For example, for the 50th percentile file, the 50th percentile January was combined with the 50th percentile February).

3. Techniques were developed for the creation of wind data (speed and direction), atmospheric pressure and cloud cover and the files were converted into a format compatible with the majority of building thermal modelling software.

  1. The outputs from the Weather Generator were compared to a set of Chartered Institution of Building Services Engineers (CIBSE) future weather files created by plotting the change in mean internal air temperature against the mean external temperature.
  1. Using these files, the impact of different levels of climate change on the built environment and building occupants was investigated. It was found that the distribution of results produced by a building thermal model running 3000 files representative of future weather (a single output of the Weather Generator) can be approximated using just five probabilistic future weather files. This has major implications when considering the uptake by industry.
  1. Psychological surveys were carried out with several engineering and architectural firms that investigated the industries attitude towards climate change, and the adequacy of current practices. Interestingly most responses regarding changes to current practice in light of climate change related to mitigation activities, such as energy saving, instead of limiting overheating risk.

Difficulties & limitations

There is a general lack of understanding surrounding the concept of probabilistic data within the building industry. Most individuals and organisations are used to deterministic data and a single result rather than distributions.

The main challenges with using the UKCP09 data to create the outputs for the project were delays in releasing the data and the lack of certain variables (wind). The updates to the Weather Generator require regular maintenance of the weather files created.

Lessons learned

The use of distributions of probabilistic data rather than single estimates highlights not only the level of uncertainty associated with the models, but also the level of risk that should be planned for when designing buildings.

How will the results be communicated to the target audience?

The outcomes of the project were communicated through a combination of academic papers, conferences, seminars and events. The building industry is using the weather files created as part of this project. To our knowledge the weather files are being used on over £3 bn worth of building projects around the UK. The outputs are also being distributed by a major building simulation software company.

This research contributed to producing probabilistic climate profiles (or ProCLip).

Find out more

· Contact details:  Tristan Kershaw, Matthew Eames and David Coley,Centre for Energy and the Environment, University of Exeter.

· The probabilistic weather files and more information from the publication can be found on the PROMETHEUS website. Or through the ARCC website.